The Final Topio Testimonial Is Out: This Time With Fraunhofer!

04—Aug—2022, by Intza Balenciaga

Meet Dr. Diego Collarana, Senior Research Engineer and Team Leader at Fraunhofer.

Dr. Diego Collarana has more than 15 years of experience conceptualising and developing enterprise information systems using artificial intelligence technologies, i.e., knowledge graphs and deep learning. His master’s studies (double degree from Universidad Politecnica de Madrid and Technische Universität Kaiserslautern) in Software Engineering gives him solid knowledge of project management, software design, and development. Furthermore, in 2018, he obtained a Ph.D. (Magna Cum Laude) at the University of Bonn in Artificial Intelligence. Dr. Collarana’s experience allows him to solve complex enterprise problems, including knowledge integration, linked data, and conversational assistants. Dr. Collarana has authored more than 40 articles in international journals and at international conferences and has served as a programme committee member at more than 20 conferences. Ms. Kathrin Lenvain, Business Development at AZO, and Intza Balenciaga, Project Manager at AZO, spoke with him about the potential of Topio and its innovation from a researcher’s perspective. Don’t miss the interview below, hot off the press!

 

1. Fraunhofer IAIS is contributing with a huge development capacity. Which latest research developments and innovations are included in the marketplace?

Fraunhofer IAIS is developing different value-added services for the Topio marketplace. One core technology asset we provide is the integration of a conversational assistant, which helps search and answer questions about the marketplace in natural language. To do this, we extract information about Topio and its data and train an AI conversational assistant to provide accurate answers to a user’s input. Another core research topic is the provision of geospatial data as geospatial knowledge graphs. Here we lift data sets offered in the Topio marketplace to a more semantically explicit and dense format. This lifting process includes expressing geospatial data and employing and aligning it with vocabularies and ontologies developed for Topio. Having data in the form of knowledge graphs eases data integration across many domains and allows for automatic reasoning to infer knowledge that is only implicitly stated in the input data. Having rich and inter-domain data opens up many possibilities for expressive and meaningful analytics to gain deep insights into geospatial data. As geospatial data is often offered on a large scale, the ability to still handle the volume and gain insights represents another research field. Here we integrated SANSA, a big data engine for the scalable processing of large-scale knowledge graphs. Finally, to recommend data sets a user might be interested in, we apply a recent AI technique called representation learning via knowledge graph embedding, which projects data sets into a metric space based on their properties and allows us to capture latent similarities, making our recommendations more meaningful

 

2. How can Topio make a difference in research? And what about the impact on those producing and willing to buy other people’s know-how/data/assets?

As an institute focused on AI and data-driven analytics solutions, we always depend on high-quality data. The means for exploring the available data sets and the intended coverage will, on the one hand, make it much easier to find the correct data to train AI methods and, on the other hand, demonstrate the impact of AI methods on real-world scenarios. Furthermore, since data offered via the Topio marketplace is always annotated with clear licensing terms, this provides a solid legal foundation that is often missing when retrieving data from the Internet.

 

3. Could you offer some examples of the innovation such marketplaces could bring to your and similar companies’ daily work?

We rely on data to prove that our techniques work as expected and can provide impact and added value. For analytics on geospatial knowledge graphs, added value is usually generated by gaining inter-domain insights. We process knowledge graphs covering various domains and derive high-quality analytics results. Compiling such a knowledge graph involves various steps. The first step is to engineer an ontology to describe the discourse’s domain(s). The next step is to find the right data sets (with compatible licences) to populate the knowledge graph with factual data. Then we transform the data sets into a suitable format, aligning overlapping data entries to ensure high-quality links between the considered domains. The Topio marketplace has value-added services for all these steps, which significantly reduce manual effort and let us focus on the AI methods, while the Topio marketplace takes care of the data market and data management.

 

Did you enjoy this interview? Anyone interested in Topio can test and explore Europe’s new geospatial data marketplace to trade and use geospatial data whenever you want and wherever you are! Contact us directly at hello@topio.market or/and sign up to the Topio mailing list to get the latest news as soon as the public beta phase starts!